Attila Bakos

1 minute ago, thephoenix said:

pffffff they'll never sort it out ?

so disappointing

Yeah but this is just a beta, and they are getting there. If I have time I'll post a ticket.

22 hours ago, BrunoCH said:

DaVinci Resolve Studio 16.1 public beta 2 released

• Support for Fujifilm F-Log colorspace

Now it's possible to use  RCM and make color grading without LUT.

 

A slight downside: it seems that the fix they implemented for the bad interpretation for Fujifilm files in v16 (bad YCbCr->RGB conversion) only works in Davinci YRGB and not in RCM.

6 minutes ago, Zeng said:

Yes, their LUT is no good. I wonder who did it for them. Don't want to be hard on Nikon, they are new to this, but still. They should hire some knowledge into their team regarding video. Neutrals not staying neutral is not that bad in my book though... At least in my attempts to match Alexa I couldn't find a way to come closer without some cast in the neutrals. Your LUT, Attila, stays true to the neutrals, but leaves too much of the Z6 N-log inherent "greenishness", at least to my liking. And I prefer even gentler  roll into the shadows.

Yes the LUTs included in my ACES packages are strictly technical, they don't include any fixes for possible color casts, it's up to the user to do that.

I still recommend my own LUTs for N-Log: http://colorizer.net/index.php?op=aces

The official LUT provided by Nikon is very amateurish, see what it does to a grey gradient:

See the ripple in the highlight rolloff? It's ugly. Let's see it with curves:

You can also see that the neutrals don't stay neutral, especially the shadows.

Btw what do you do with the gamut? Vlog L primaries are not published, do you use an estimation?

15 minutes ago, Inazuma said:

That may be so, but the xt20 shared the same internals as the xt2 and had noticeably worse rolling shutter. 

The readout was different though. The X-T2 had a 1.17x crop with full readout, the X-T20 had no crop but line skipping.

On 7/22/2019 at 12:49 AM, andrgl said:

@mercer

You can generate a LUT from Lightroom grades with this program:

https://generator.iwltbap.com/

Kinda convoluted. You make a HALD file, apply you Lightroom preset, save it, then use the program to export a CUBE file.

The LUT only affects color, so no texture, clarity, denoise, etc.

If you don't want to install anything, I created an online version that does the same thing: http://colorizer.net/index.php?op=lightroom

I just updated to Resolve 16 beta and the issue seems to be fixed. All Fujifilm files including HLG are interpreted with the correct matrix. The changelog mentions Fujifilm files but the fix seems to be general, it's seems that the matrix coefficients tag is now respected. The latest Premiere still has issues though.

Just now, thephoenix said:

thanks, but personaly i don't know anyone that could actualy build the cable even with wiring diagram

i am not even sure that what is working on the crane will work with an other gimbal. not even talking of the nucleus nano

Sorry it wasn't a reply to your issue, I just thought I mention it as we are talking about cables.

I friend of mine has a bit of electrical engineering knowledge, he created a cable that allows me to control the X-T3's shutter and focus from the Zhiyun Crane Plus. It works on the Zhiyun Crane 2 as well. He won't create another one but I can ask for the wiring diagram if anyone's interested. It needs an optocoupler and a couple of resistors, so a part of the cable will be thick, but it's not a real problem.

36 minutes ago, The ghost of squig said:

Yeah it's the 95MB/s write card. Interestingly, Sandisk lowered the specified write speed on the 170MB/s read cards to 90MB/s write. 400Mbits/s write is only 50/MB/s so both cards have plenty of headroom. The faster read speed will be welcome ?

The headroom is smaller, the X-T3 can't write this card at full speed. It's somewhere between 60-70 MB/s, but it's still enough. It all depends on the actual body, afaik the GH5 can't even write at 50 MB/s to this card.

17 minutes ago, jagnje said:

what sd cards do you guys use? 

I use Lexar 2000x cards, but there was a discount for these Sandisks recently that I couldn't resist: SanDisk Extreme PRO 256 GB SDXC Memory Card, Up to 170 MB/s, Class 10, U3, V30

They seem work well for 400Mbit/s recording so far and they are way cheaper than my Lexars.

9 hours ago, thephoenix said:

i don't really understand why this app is for android systems as most of editors works on computers when it comes to video.

FYI:

On 5/20/2019 at 8:31 AM, kinoseed said:

You can use the app via any modern browser or Android device.
Web: https://kinoseed.com/

5 hours ago, thephoenix said:

didn't know there was the same issue with hlg.

well i use the ninja with the xt3 so internal files are only here as backup files.

so to fix my issue i have 2 options, 1 is convert all my files 2 buy your lut package

Yes, but hopefully they fixed all these issues in 16. Will report back once I can test it. I'm still on 15.

22 minutes ago, thephoenix said:

well i re watched the video and ok it is clear when you apply the lut, but my concern is that you don't use rcm. it's funny how few people use rcm when you know how powerfull and usefull it is.

i think i understood anyway why my test still had the color issue. i didn't checked the conversion to 709 in shutter encoder.

i could use ffmpeg yes but

1 i have over 100 files to convert so i don't have 2 days to spare to conversion

2 shutter encoder does use ffmpeg

good news is now problem is fixed in 16b3 and i shoot hlg so no issue with it

If they fixed it then that's the best solution FYI hlg was also interpreted incorrectly, I had to create a different matrix correction for that issue. I hope they fixed it as well.
I don't have the beta installed but it would be interesting to see if they only fixed this for Fujifilm files (as the changelog dictates), or in general. The issue is not specifically about Fujifilm files, but I guess Fuji people were the only one complaining.

13 hours ago, bolex said:

Great work!!! Have you notice macroblocking issue like in xt2? Is xt3 log easier to work with then xt2's?

Thanks! Some people reported about banding and magenta casts, but luckily I haven't seen any of those yet. 400Mbps and 10bit helps a lot, the quality of F-Log compared to the X-T2 is night and day. I can't say it's easier or harder to work with. Colors changed a bit, but it's a slight change. Gamma and color space is the same as before.

4 minutes ago, androidlad said:

Looks like the LUT has a bit earlier/harsher highlight clipping compared to native film simulations? (Judging from the highlight patch around the lady's collarbone)

At first I also thought that I messed up something, but I didn't. Unfortunately there were changes in light conditions, and I only noticed it at home. There was a tree above her that allowed a bit more light to fall to the collarbone in the F-Log shot. Because of this, please don't judge rolloff from that shot. It should be the same. Good eyes

Finally finished this project:

More info: http://colorizer.net/index.php?op=xt3

12 hours ago, thephoenix said:

to me the real question is "will it still be there after transcoding to prores or dnxhr" because you don't want to work with h265 files directly in resolve. or you have to create proxy files.

did a test and i had a clear difference in red

If you convert your clips anyway, why don't you convert the matrix? It's a bulletproof solution to this issue which I described in my video I recommended you to watch.
Or if you prefer a written version, here it is at the bottom of the page: http://colorizer.net/index.php?op=technical
It uses FFMPEG.

28 minutes ago, thephoenix said:

you clearly see a difference on my pic anyway. so looks like problem it not solved.

one question.

problem is with fuji flog internal files.

it means ninja files are not affected right ?

to correct it do we have to apply the fuji lut flog to rec709 ? or do we have to always apply this lut to files coming from the fuji and also the ninja ?

because when you mix sources it is a bit a tricky to apply the lut only to some rushes.

or maybe you can convert fuji internal files to prores or dnxhd applying the lut ? (shutter encoder can do that)

 

Unfortunately I don't have time to go into the details right now, but I recommend you watch the video in the thread I linked earlier (previous page), it has the answers to most of your questions.
 

34 minutes ago, keessie65 said:

Thanks Attila, I used F-Log for this one and graded with Venice Cine lut; https://www.sierracreativeco.com.

The oval at 2:10 is from the lens, colours from the lut.

I like the oval, but the color transitions are a bit harsh around the sun. Nothing serious though, nice colors overall.

40 minutes ago, androidlad said:

Strictly speaking it's full range 0-1023. But Fuji encodes nothing between 0-95.

Yes, that's correct.

30 minutes ago, BrunoCH said:

The most accurate way will be when Resolve will have an F-log preset for rushes. 

For what exactly do we need more accuracy? We know all about F-Log, all the info is there in the data sheet provided by Fujifilm.

2 minutes ago, BrunoCH said:

I think fuji log use full range into the limit of video range (Panasonic log don't use full range into video range) 

Fuji log uses 95-1023, because of the gamma curve. That's still a wider range than video range (64-940).

39 minutes ago, BrunoCH said:

Full data is about still photo, like jpg dpx etc (image sequence); RGB file.

This is a bit oversimplified, a lot of cameras use the full range in their video files. Fuji is one of them.

20 minutes ago, thephoenix said:

ok. can you tell me more about full data level and why is it not set to full by default ?

just discovered that it was set to auto and when setting to full then image changes and seems to be flatter than without (flog)

ProRes is always interpreted as video range, even if it's a full range recording. I'm not even sure ProRes has flags for setting the data levels at all. When you set it manually to full range it becomes flatter, that's normal.

Just now, thephoenix said:

well... since i have the free version h265 will not open in resolve

can convert to dnxhr using shutter encoder but will it change something ?

Don't bother with converting, we need to see how Resolve handles the original files. Maybe someone else can help us out.